Irs Wtc2020

Intelligent Reflecting Surface Aided Wireless
Communication: Opportunities and Challenges
Rui Zhang
Professor, National University of Singapore

IEEE Fellow, Distinguished Lecturer
Clarivate Analytics Highly Cited Researcher
(e-mail: elezhang@nus.edu.sg)
WTC Online Seminar

Sept 23, 2020
WTC Online Seminar 2020 1
Rui Zhang, National University of Singapore
Outline
 Introduction of Intelligent Reflecting Surface (IRS)
 Motivation
 Hardware architecture
 Reflection and channel models
 Main functions and applications
 Comparison with existing wireless technologies
 Communication Design Challenges

 IRS reflection optimization
 IRS channel estimation
 IRS deployment
 Other Applications/Extensions
 Conclusion and Future Work

Outline
 Motivation

 IRS deployment
 Conclusions and Future Work

Motivation Rui Zhang, National University of Singapore
Have We Reached Shannon’s Capacity Limit?

𝐻𝐻𝑃𝑃
𝐶𝐶 = log 1 + 2
𝜎𝜎
 Yes, also No (as wireless channel H is still random and uncontrolled)
 Can we make H arbitrarily large, say from H<<1 to H→1?
 Can we make H less random, e.g., from Rayleigh fading to Rician fading?
 Existing wireless technologies (beamforming, power control, adaptive
modulation, etc.) only adapt to H, but have no control over it
 How to break this ultimate barrier to achieving ultra-high capacity and ultra-
high reliability in future wireless communications (e.g., 6G)?
 Promising new paradigm: Smart and Reconfigurable Wireless Environment
 Key enabling technology: Intelligent Reflecting Surface (IRS)
 Other nomenclature: reconfigurable intelligent surface (RIS), software
controlled metasurface, passive intelligent mirror, smart reflect array, ….
Hardware architecture Rui Zhang, National University of Singapore
What is IRS?
Communicate with other nodes in the

network (IRS, BS, terminal) and adjust the
reflecting elements dynamically
 A digitally-controlled metasurface with massive low-cost passive reflecting

elements (each able to induce an amplitude/phase change in the incident signal)
 Low energy consumption (without the use of any transmit RF chains), high
spectral efficiency (full-duplex, noiseless reflection)
Q. Wu and R. Zhang, “Towards Smart and Reconfigurable Environment: Intelligent Reflecting Surface Aided Wireless
Networks,” IEEE Communications Magazine, January 2020.

Reflection model Rui Zhang, National University of Singapore
IRS: Reflection Model
 Baseband equivalent signal model at each IRS element
𝑛𝑛 = 1, … , 𝑁𝑁
where : reflection amplitude

: phase shift
𝑁𝑁 : No. of elements
: Absorption
: Full reflection
 In practice, both amplitude and phase shift need to be discretized

Channel model Rui Zhang, National University of Singapore
IRS: Channel Model

 Baseband equivalent channel model (narrow-band)
 Assume isotropic reflection, and no mutual
coupling among reflecting elements
𝑑𝑑1𝑛𝑛 element 𝑛𝑛 𝑑𝑑2𝑛𝑛

complex channel complex reflection ℎ𝑛𝑛 𝑔𝑔𝑛𝑛
coefficients coefficient
𝑥𝑥 𝑦𝑦
 Product-distance path loss model  𝑥𝑥: transmitted signal
 𝑦𝑦: received signal
 ℎ𝑛𝑛: first link channel
 g𝑛𝑛: second link channel
 Extendible to wide-band channel, with IRS frequency-flat reflection only

Channel model Rui Zhang, National University of Singapore
IRS Path Loss Model: Product Distance or Sum Distance?

 Product-distance path loss model
𝑑𝑑1𝑛𝑛 element 𝑛𝑛 𝑑𝑑2𝑛𝑛

ℎ𝑛𝑛 𝑔𝑔𝑛𝑛
 Sum-distance path loss model
𝑥𝑥 𝑦𝑦
 Applies to free-space propagation and

infinitely large perfect electric conductor
(PEC) only
 Not applicable to IRS with finite-size
elements
IRS functions Rui Zhang, National University of Singapore
Main Functions of IRS in Wireless Communication
Channel reconfiguration
Passive beamforming
Interference nulling/cancelation
…
Q. Wu and R. Zhang, “Towards Smart and Reconfigurable Environment: Intelligent Reflecting Surface
Aided Wireless Networks,” IEEE Communications Magazine, January 2020.

IRS for Channel Reconfiguration
 Create virtual LoS link by smart

reflection to bypass obstacle
 Coverage extension for mmWave
 Add extra signal path toward desired
direction
 Improve channel rank and thus
spatial multiplexing gain
 Refine channel statistics/distribution
 Transform Rayleigh/fast fading to
Rician/slow fading for ultra-high
reliability

IRS-enabled Passive Beamforming
 3D passive beamforming for broadcasting/multicasting

 Enhance signal power/SNR at “cell edge” or “hot spot”
 Boost network capacity without additional signal transmission
IRS-assisted Interference Nulling/Cancelation
 Enhance desired signal power

while nulling the reflected
interference
 Alternatively, tune the reflected
interference to cancel the direct
interference (more challenging to
implement)
 Both improve cell-edge user’s SINR
 Create a “signal hotspot” as well
as “interference-free zone” in the
vicinity of IRS

IRS applications Rui Zhang, National University of Singapore
IRS Applications for 5G/6G
Q. Wu, S. Zhang, B. Zheng, C. You, and R. Zhang, “Intelligent reflecting surface aided wireless
communications: a tutorial,” submitted to IEEE Transactions on Communications (Invited Paper)
IRS applications Rui Zhang, National University of Singapore
Related Industry Initiatives
 Metawave
 Passive reflector/relay
(ECHO) 28 GHz-band
 Greenerwave
 Reconfigurable binary
metasurface
 Pivotal Commware
 Holographic beam
forming

Comparison with existing technologies Rui Zhang, National University of Singapore
IRS vs Active Relay/Small Cell/DAS/Cell-free MIMO

mmWave BS
DAS
Relay
UDN
 Network with active BS/AP/relay only  Hybrid active-passive network: fewer BSs
 High cost, high energy consumption with many passive IRSs
 Backhaul issue  Low cost, low energy consumption
 Complicated interference management  Low-rate wireless backhaul suffices (for
control link only)
 Low spectral efficiency due to half duplex
(full-duplex radio needs costly self-  Local coverage only without the need of
interference cancelation) inter-IRS interference management
 Full duplex without self-interference
Wireless Network with IRS vs w/o IRS: System Setup
 A fundamental question for IRS:

 Can large-scale deployment of IRSs
provide cost-effective & sustainable
capacity growth in wireless network?
 New hybrid active/passive

network with IRS
 Randomly distributed active BSs and
passive IRSs subjected to inter-cell
interference
 Characterize network coverage probability and spatial throughput in terms of key
system parameters including BS/IRS densities and network loading factor
 Analytical framework based on stochastic geometry

J. Lyu and R. Zhang, “Hybrid active/passive wireless network aided by intelligent reflecting surface: System modeling and
performance analysis,” submitted to IEEE Transactions on Wireless Communications. https://arxiv.org/abs/2004.13318

Wireless Network with IRS vs w/o IRS: Simulation Results

Loading factor
 Spatial throughput ν subject to total
BS/IRS cost C
 IRS/BS density ratio:
 Cost of each BS: , BS/IRS cost ratio:
 Total cost C per m2 in the hybrid network:
 ν versus ζ under given total cost C:

 Optimal ratio to achieve maximum exists
 Significantly outperforms BS-only network (ζ=0)
 ζ too large: no enough signal power for
effective IRS reflection and beamforming
 ν versus C under given density ratio ζ :

 BS-only network: ν first increases then
decreases due to more severe
interference than improved signal power
 Hybrid network with optimal :
maximum increases almost linearly
with C → cost-effective and sustainable
capacity growth

IRS vs Massive MIMO/Active Large Surface

IRS
Massive MIMO IRS-aided Small MIMO

(Non-scalable with increased frequency) (Scalable at any frequency)
 More RF chains needed for more active  No RF chains needed for IRS due
elements used to passive reflection only
 Increased energy consumption, hardware  Low energy consumption, scalable
cost, and processing complexity at higher cost/complexity
frequencies (mmWave, THz)  Compatible with cellular/WiFi and
can be densely deployed
IRS vs Massive MIMO: Simulation Setup

 IRS-aided small MIMO vs massive MIMO without IRS, both TDD-based
 8 users randomly distributed within 60 m from AP and 8 users randomly
distributed within 6 m from IRS
 Transmission protocol for IRS-aided small MIMO
1) all users send orthogonal pilot signals concurrently
2) AP and IRS estimate AP-user and IRS-user channels, respectively
3) AP starts to transmit data to users and in the meanwhile sends its estimated AP-user channels to
IRS controller via a separate control link
4) IRS controller sends optimized transmit beamforming vectors to AP and sets its phase shifts
accordingly
5) AP and IRS start to transmit data to users jointly
 Two transmission phases for IRS-aided small MIMO
 Small MIMO transmission of duration,
 IRS-aided joint MIMO transmission of duration,
 Delay ratio:
Q. Wu and R. Zhang, “Intelligent reflecting surface enhanced wireless network via joint active and
passive beamforming,” IEEE Transactions on Wireless Communications, November 2019.
IRS vs Massive MIMO: Simulation Results

7.5
M=20, N=80
M=20, N=120 7
7 M=20
M=40
M=50 6.5
6 M=60
Achievable max-min rate (bps/Hz)

Achievable max-min rate (bps/Hz)
M=20, N=80
5.5
M=20, N=120
M=20
4 M=40
5
M=50
M=60
3 4.5
4
2
3.5
4 6 8 10 12 14 16 0 0.05 0.1 0.15 0.2 0.25 0.3
AP transmit power (dBm) Delay ratio,
 M: # of active antennas at AP  This also holds considering delay due to

 N: # of reflecting units at IRS
 Passive IRS helps reduce # of active IRS channel estimation (compare M=20
antennas (see M=20 with N=80 vs M=50 with N=80 vs M=40 without IRS)
without IRS)
Outline
 Motivation

 IRS deployment
 Conclusion and Future Work

IRS reflection optimization Rui Zhang, National University of Singapore
Joint Active and Passive Beamforming: Single-user Case
 AP: active (transmit) beamforming

 IRS: passive (reflect) beamforming with maximum reflection amplitude ( )
 Objective: maximize the received signal power via joint transmit and reflect
beamforming optimization
 Establish a local “signal hotspot” in the vicinity of IRS
 Received SNR scaling order: O(N2)
 Thanks to the dual role of “receive” and “reflect” (full-duplex, noise-free), in contrast to O(N) of massive
MIMO (limited by sum-power constraint at Tx), and O(N) of MIMO AF relay (due to relay noise)
 Hold even for practical IRS with discrete phase shifts
Q. Wu and R. Zhang, “Intelligent reflecting surface enhanced wireless network: joint active and passive beamforming design,” in Proc. IEEE
GLOBECOM, 2018.
Q. Wu and R. Zhang, “Beamforming optimization for intelligent reflecting surface with discrete phase shifts,” in Proc. IEEE ICASSP, 2019.
Minimum AP transmit Power vs AP-user Distance

 Transmit beamforming: 𝒘𝒘
 Reflect beamforming: ϴ
Problem formulation (suboptimal solutions

obtained via SDR or alternating optimization)
Proposed
Simulation setup
 Significant power saving with IRS (vs w/o IRS)

 Performance gain of joint transmit and reflect
beamforming design (vs AP-user MRT or AP-IRS
MRT benchmarks)

Minimum AP transmit Power vs No. of IRS Elements (1)

d = 50 m
 SNR scaling law O(N2) for

sufficiently large N, near IRS
Increasing N from 30 to 60
results in 6 dB power gain/saving
d = 41 m d = 15 m

Minimum AP transmit Power vs No. of IRS Elements (2)
 Finite number of levels

of phase-shift: 2𝑏𝑏
 b: No. of phase-
control bits
 Power loss of using b

phase-control bits
 Suboptimal solution obtained via uniformly quantizing the continuous-phase solution

 SNR scaling law, i.e., O(N2), still holds with finite-level phase shifters
 IRS with 1-bit (2-bit) phase-shifters suffers a power loss of 3.9 dB (0.9 dB)
Q. Wu and R. Zhang, “Beamforming optimization for wireless network aided by intelligent reflecting
surface with discrete phase shifts,” IEEE Trans. Commun., March 2020.
Joint Active and Passive Beamforming: Multi-user Case
 Objective: minimize total transmit power at the AP subject to individual user SINR
constraints via joint transmit and reflect beamforming optimization
 Establish a “signal hotspot” as well as an “interference-free zone” near IRS
 Reflect beamforming by IRS
 help enhance SINR of the users near IRS
 Enable more flexible AP transmit beamforming toward users outside IRS coverage
 Thereby improve the overall network SINR performance
Q. Wu and R. Zhang, “Intelligent reflecting surface enhanced wireless network via joint active and
passive beamforming,” IEEE Transactions on Wireless Communications, November 2019.
Minimum AP transmit Power vs User SINR Target

 Multi-user problem formulation
Simulation setup
 Significant power saving at AP with vs. w/o IRS

 Special two-user case: one user near IRS  Two algorithms proposed: Alternating optimization
and the other user far from IRS or two-stage algorithm (details omitted here)

Signal/Interference Power vs SINR Target
User 1: far from IRS, served by AP User 2: near IRS, served by both AP
transmit BF only BF and IRS (reflect) BF
 IRS provides not only signal power gain, but also interference mitigation gain for
near user (user 2), which also benefits for far user (user 1)

IRS: Practical Phase Shift Model

 Reflection coefficient at each IRS element
where : element impedance

: free space impedance
 Ideal model: maximum (unit) reflection amplitude

regardless of phase shift (widely used in the literature, but
difficult to realize in practice)
 Practical model: minimum amplitude occurs near zero
phase shift and approaches unity (maximum) at
 Reason: when phase shift approaches zero, image currents
become in-phase with reflecting element currents, thus
more energy loss and hence low reflection amplitude
 cannot be zero in practice (varactor diodes )
 Implication: IRS passive beamforming needs to balance
between reflection amplitude and phase alignment
S. Abeywickrama, R. Zhang, Q. Wu, and C. Yuen, “Intelligent reflecting surface: practical phase shift model and
beamforming optimization,” IEEE Transactions on Communications, Early Access

Analytical IRS Model with Phase Shift Dependent Amplitude
 Generally applicable to a variety

of semiconductor devices used
for implementing the IRS
 Reflection coefficient at each IRS element

𝑛𝑛 = 1, … , 𝑁𝑁
where
phase shift:
reflection amplitude:

Achievable Rate with Practical vs Ideal Model
 Single-user problem formulation

 Transmit beamforming: 𝐰𝐰; Reflect beamforming: 𝐯𝐯
 Suboptimal solution obtained via alternating
optimization (AO)

Selected Work on IRS Reflection Optimization

 Power minimization
• Q. Wu and R. Zhang, “Intelligent reflecting surface enhanced wireless network via joint active and passive
beamforming,” IEEE Transactions on Wireless Communications, November 2019.
 Energy efficiency maximization
• C. Huang, A. Zappone, G. C. Alexandropoulos, M. Debbah, and C. Yuen, “Reconfigurable intelligent surfaces for
energy efficiency in wireless communication,” IEEE Trans. Wireless Commun., Aug. 2019.
 Rate maximization
• H. Guo, Y.-C. Liang, J. Chen, and E. G. Larsson, “Weighted sum-rate optimization for intelligent reflecting
surface enhanced wireless networks,” arXiv preprint arXiv:1905.07920, 2019.
 IRS-aided OFDM
• Y. Yang, B. Zheng, S. Zhang, and R. Zhang, “Intelligent reflecting surface meets OFDM: protocol design and rate
maximization,” IEEE Transactions on Communications, July 2020.
 IRS-aided MIMO
• S. Zhang and R. Zhang, “Capacity characterization for intelligent reflecting surface aided MIMO
communication,” IEEE Journal on Selected Areas in Communications, August 2020.
• C. Pan, H. Ren, K. Wang, W. Xu, M. Elkashlan, A. Nallanathan, and L. Hanzo, “Multicell MIMO communications
relying on intelligent reflecting surface,” arXiv preprint arXiv:1907.10864, 2019.
 IRS optimization with discrete phase/hardware imperfection
• Q. Wu and R. Zhang, “Beamforming optimization for wireless network aided by intelligent reflecting surface
with discrete phase shifts,” IEEE Transactions on Communications, March 2020.
• S. Abeywickrama, R. Zhang, Q. Wu, and C. Yuen, “Intelligent reflecting surface: practical phase shift model and
beamforming optimization,” IEEE Transactions on Communications, early access.
IRS channel estimation Rui Zhang, National University of Singapore
IRS Channel Estimation

 Channel estimation in IRS-aided wireless network
 BS-user links (existing): estimated by conventional methods and switching off IRS
 BS-IRS link (new): quasi-static with fixed BS and IRS
 IRS-user links (new): vary with user location, needs to be estimated in real time
 Main challenges in IRS channel acquisition

 Passive IRS (no Tx RF chains): IRS cannot send pilot signals for channel estimation
 Large number of extra channel coefficients: O(MN+NK). M: # of BS antennas; N: # of IRS
elements; K: # of users
 IRS performance gains critically depend on the CSI in general
 Three general approaches

 Equip IRS with active elements/sensors (semi-passive IRS)
 Estimate BS-IRS-user cascaded channels (fully-passive IRS)
 Beam searching without explicit channel estimation (codebook-based)
Q. Wu and R. Zhang, “Towards Smart and Reconfigurable Environment: Intelligent Reflecting Surface
Aided Wireless Networks,” IEEE Communications Magazine, January 2020.

Channel Estimation with Semi-Passive IRS
 IRS with active elements

/sensing devices
 Channels estimated by IRS
leveraging TDD and
channel reciprocity
 Signal processing required
for reconstructing the IRS-
BS/user channels with
limited/low-cost sensors
Channel Estimation with Fully-Passive IRS
 IRS w/o active elements

and sensing devices
 More challenging case (as
compared to semi-passive
IRS)
 Cascaded channel (BS-IRS-
user) estimation by varying
IRS reflection and exploiting
the static/sparse BS-IRS
channel (common for all
users): an active area of
research!
Cascaded Channel Estimation: Useful Techniques

 Orthogonal IRS reflection design for training (e.g., DFT/Hadamard matrix) to resolve
different user channels, nearly orthogonal for practical discrete phase
 IRS elements grouping: divide IRS elements into adjacent groups with common phase
shift per group to exploit channel correlation and reduce training/passive
beamforming complexity: O(N)→O(M), with M groups
 Common BS-IRS channel exploitation: estimate first the cascaded channel for a
reference user and then resolve those of the other users more efficiently
 Progressive channel estimation: start with coarse channel estimation for large-size
groups (for initial connection) and gradually refine channel estimation for smaller-size
groups to improve beamforming gains (for high-rate data transmission)

Beam Searching without Explicit Channel Estimation

 Codebook-based IRS beam searching
 Multi-beam training (new): divide IRS reflecting elements into sub-arrays and
design their simultaneous multi-beam steering over time
 More efficient than conventional single-beam training
C. You, B. Zheng, and R. Zhang, “Fast beam training for IRS-assisted multiuser communications,”
IEEE Wireless Communications Letters, Early Access.
Selected Work on IRS Channel Estimation

 Semi-passive IRS
• A. Taha, M. Alrabeiah, and A. Alkhateeb, “Enabling large intelligent surfaces with compressive sensing and
deep learning,” arXiv preprint arXiv:1904.10136, 2019.
 Fully-passive IRS
 IRS elements grouping and channel estimation in OFDM/OFDMA
• B. Zheng and R. Zhang, “Intelligent reflecting surface enhanced OFDM: channel estimation and reflection
optimization,” IEEE Wireless Communications Letters, April 2020.
• B. Zheng, C. You, and R. Zhang, “Intelligent reflecting surface assisted multi-user OFDMA: channel
estimation and training design,” IEEE Transactions on Wireless Communications, Early Access.
 IRS channel estimation in massive MIMO
• Z.-Q. He and X. Yuan, “Cascaded channel estimation for large intelligent metasurface assisted massive
MIMO,” IEEE Wireless Commun. Lett., Feb. 2020.
• Z. Wang, L. Liu, and S. Cui, “Channel estimation for intelligent reflecting surface assisted multiuser
communications: Framework, algorithms, and analysis,” IEEE Trans. Wireless Commun., Early Access.
 Progressive IRS channel estimation with discrete phase
• C. You, B. Zheng, and R. Zhang, “Channel estimation and passive beamforming for intelligent reflecting
surface: discrete phase shift and progressive refinement,” IEEE JSAC, Early Access.
 IRS beamforming searching
• C. You, B. Zheng, and R. Zhang, “Fast beam training for IRS-assisted multiuser communications,” IEEE Wireless
Communications Letters, Early Access.
IRS deployment Rui Zhang, National University of Singapore
IRS Deployment: Single-User Case

 Consider one single IRS with 𝑁𝑁 reflecting elements, assume no direct link
60
50
Received SNR (dB)

40
IRS near user IRS near AP
30
 Double path loss (product distance) 20
 Received SNR: 10
0 100 200 300 400 500
IRS-user horizontal distance, d (m)
 IRS-AP distance generally increases as the user-IRS distance decreases

 Optimal deployment strategy to maximize received SNR (minimize product distance):
𝑑𝑑 = 0 (IRS near user) or 𝑑𝑑 = 𝐷𝐷 (IRS near AP), different from active relay (d=D/2)
 Maximum received SNR:

IRS Deployment: Cooperative/Double IRSs

 Given 𝑁𝑁 IRS reflecting elements, forming them as two cooperative IRSs
90
Single IRS
80 Two cooperative IRSs
70
Received SNR (dB)

60
50
 Pros: Cooperative beamforming gain

40
with order (vs. for 30
single-IRS case)
0 200 400 600 800 1000
Total number of IRS reflecting elements, N
 Cons: Double reflection, more path loss

 Two cooperative IRSs outperform one
 Received SNR: single IRS if
Y. Han, S. Zhang, L. Duan, and R. Zhang, “Cooperative double-IRS aided communication: beamforming
design and power scaling,” IEEE Wireless Communications Letters, August 2020.
General Channel Case: Single or Double IRSs?

 Consider multi-antenna BS, all single- and double-reflection links
 Double IRS generally outperforms single IRS at any SNR (theoretically proved if
K=1 or single-user case)
 Double IRS also provides larger effective channel rank than single IRS (thus,
higher spatial multiplexing gain)
B. Zheng, C. You, and R. Zhang, “Double-IRS assisted multi-user MIMO: cooperative passive beamforming
design,” submitted IEEE Transactions on Wireless Communications, available at arXiv:2008.13701
IRS Deployment: Centralized IRS or Distributed IRSs?

 Centralized deployment: Deploy all  Distributed deployment: Deploy 𝑁𝑁𝑘𝑘
𝑁𝑁 reflecting elements near the AP elements near each user 𝑘𝑘, ∑𝐾𝐾
𝑘𝑘=1 𝑁𝑁𝑘𝑘 = 𝑁𝑁
 Each user is only served by its nearby IRS

 Both users are served by the centralized
with 𝑁𝑁𝑘𝑘 (𝑁𝑁𝑘𝑘 < 𝑁𝑁) elements
IRS with 𝑁𝑁 elements
 Pros: Beamforming gain is maximized for
 Pros: Larger beamforming gain
each user
 Cons: Beamforming gain needs to be
 Cons: Only beamforming gain for
shared by the two users
each user
S. Zhang and R. Zhang, “Intelligent reflecting surface aided multiple access: capacity region and
deployment strategy,” IEEE SPAWC, 2020. Available: https://arxiv.org/abs/2002.07091

IRS Deployment: Centralized IRS or Distributed IRSs?

 Capacity region comparison for a two-user multiple access channel (MAC):
 Under symmetric channel condition, centralized deployment outperforms distributed

deployment in terms of capacity region, due to more flexibility in trading off users’
individual reflected channels as compared to distributed IRSs
 In practice, other factors need to be taken into account, such as space constraint, LoS
availability, etc.
S. Zhang and R. Zhang, “Intelligent reflecting surface aided multiple access: capacity region and
deployment strategy,” IEEE SPAWC, 2020. Available: https://arxiv.org/abs/2002.07091
Selected Work on IRS Deployment

 Double/cooperative IRS
• Y. Han, S. Zhang, L. Duan, and R. Zhang, “Cooperative double-IRS aided communication: beamforming design
and power scaling,” IEEE Wireless Communications Letters, August 2020.
• B. Zheng, C. You, and R. Zhang, “Double-IRS assisted multi-user MIMO: cooperative passive beamforming
design,” submitted IEEE Transactions on Wireless Communications, available at arXiv:2008.13701
• C. You, B. Zheng, and R. Zhang, “Wireless communication via double IRS: channel estimation and passive
beamforming designs,” submitted to IEEE Wireless Communications Letters. https://arxiv.org/abs/2008.11439
 Centralized vs distributed IRS
• S. Zhang and R. Zhang, “Intelligent reflecting surface aided multiple access: capacity region and deployment
strategy,” IEEE SPAWC, 2020. Available: https://arxiv.org/abs/2002.07091
• S. Zhang and R. Zhang, “Intelligent reflecting surface aided multi-user communication: capacity region and
deployment strategy,” submitted to IEEE Transactions on Communications. https://arxiv.org/abs/2009.02324
 IRS deployment in large network based on stochastic geometry
• J. Lyu and R. Zhang, “Spatial throughput characterization for intelligent reflecting surface aided multiuser
system,” IEEE Wireless Communications Letters, June 2020.
• M. A. Kishk and M.-S. Alouini, “Exploiting randomly-located blockages for large-scale deployment of intelligent
surfaces,” arXiv preprint arXiv:2001.10766, 2020.
 IRS deployment based on machine learning
• X. Liu, Y. Liu, Y. Chen, and H. V. Poor, “RIS enhanced massive non-orthogonal multiple access networks:
Deploymentand passive beamforming design,” arXiv preprint arXiv: 2001.10363, 2020.

Outline
 Motivation

 IRS deployment
 Conclusions and Future Work

IRS-aided Multiple Access
 B. Zheng, Q. Wu, and R. Zhang, “Intelligent reflecting surface-assisted multiple access

with user pairing: NOMA or OMA?” IEEE Communications Letters, vol. 24, no. 4, pp.
753-757, April 2020.
 Y. Yang, S. Zhang, and R. Zhang, “IRS-enhanced OFDMA: Joint resource allocation and
passive beamforming optimization,” IEEE Wireless Communications Letters, vol. 9, no.
6, pp. 760-764, June 2020.
 X. Guan, Q. Wu, and R. Zhang, “Joint power control and passive beamforming in IRS-
assisted spectrum sharing,” IEEE Communications Letters, vol. 24, no. 7, pp. 1553-
1557, July 2020.

IRS-aided Multiple Access Rui Zhang, National University of Singapore
IRS-aided Multiple Access: OMA vs NOMA

 Conventional multiple access without IRS
 NOMA is always superior to OMA (TDMA/FDMA)
 TDMA and FDMA achieve the same theoretical performance
 IRS-aided multiple access
 Setup: two users share two given adjacent
time-frequency resource blocks (RBs)
 TDMA is always superior to the FDMA (due to
passive IRS reflection that can be time-
selective, but cannot be frequency-selective)
 NOMA is always superior to the FDMA (due to
IRS controller
...
...
IRS
the higher spectrum efficiency of NOMA with s1
hr
gH
1
any passive IRS reflection ) s2 hd ,1

U1
AP
 NOMA may perform worse than TDMA for hd ,2
g 2H
near-IRS users with symmetric rates U2

IRS-aided Multiple Access: OMA versus NOMA

 Problem formulation for NOMA:  Problem formulation for FDMA:
• 𝝀𝝀𝒊𝒊 (𝜽𝜽) :Channel power  Frequency-flat IRS reflection
gain of user 𝑖𝑖
• 𝑃𝑃𝑖𝑖 : Transmit power for
user 𝑖𝑖
• 𝜎𝜎 2 : Noise variance
 Problem formulation for TDMA:
 Time-selective IRS
reflection
 The user decoding order of NOMA can be
permuted by adjusting the IRS reflection
 Asymmetric user deployment: one near-IRS user and one

 Symmetric user deployment: two near-IRS users far-IRS user

IRS-enhanced OFDMA: Dynamic Passive Beamforming
 BS & 𝐾𝐾 Users: single-antenna

 Resource allocation of 𝑁𝑁 sub-bands over 𝑄𝑄 time slots
 𝑀𝑀 dynamic IRS passive beamforming reflection coefficients 𝜙𝜙𝑞𝑞,𝑚𝑚
 BS-user direct link: 𝒉𝒉𝑑𝑑𝑘𝑘 with 𝐿𝐿0,𝑘𝑘 taps
 BS-IRS-user link: 𝒉𝒉𝑟𝑟𝑘𝑘,𝑞𝑞 = 𝑽𝑽𝑘𝑘 𝝓𝝓𝑞𝑞 with 𝐿𝐿1 + 𝐿𝐿2,𝑘𝑘 − 1 taps
5
Dynamic Beamforming
Fixed Beamforming
4.5
Random Phase
Without IRS
4
Common Rate (bps/Hz)
3.5
2.5
0 10 20 30 40 50
Number of Reflecting Elements, M

IRS-aided Spectrum Sharing: A Cognitive Radio Approach

IRS IRS IRS IRS
P1(PR) S1(SR) P1(PT) S1(ST) P1(PT) S1(SR) P1(PR) S1(ST)
P2(PT) S2(ST) P2(PR) S2(SR) P2(PR) S2(ST) P2(PT) S2(SR)
Scenario (a) Scenario (b) Scenario (c) Scenario (d)
 Challenging scenarios (limited SU rate in conventional CR system without IRS):

 Strong cross-link interference when the SU is located nearby the PU
 The asymmetric interference scenarios (c, d) is even more challenging than symmetric ones (a, b)
 Objective: maximize the SU rate via joint power control and passive beamforming
 Problem formulation:
log (1+γ s )
2 2
max ps v H h srs + hss ps v H h srs + hss
ps , v =γs = 2
，γ p 2
，
s.t. γ p ≥ γ th , p p v h prs + hps + σ s
H 2
p p v h prs + hps + σ s
H 2
ps ≥ Pmax , vn =
1, n =
1,..., N . hirj diag ( h rjH ) h*ir ,i ∈ { p, s} , j ∈ { p, s}
=
which is solved by using the alternating optimization technique

IRS-aided Spectrum Sharing: Simulation Results

IRS IRS IRS IRS
P1(PR) S1(SR) P1(PT) S1(ST) P1(PT) S1(SR) P1(PR) S1(ST)
P2(PT) S2(ST) P2(PR) S2(SR) P2(PR) S2(ST) P2(PT) S2(SR)
Scenario (a) Scenario (b) Scenario (c) Scenario (d)
Scenario (a) Scenario (c) Scenario (d)
 IRS significantly improves the secondary user rates, especially in scenario (d),
where ST and PR are near the IRS
IRS-aided PHY Security
 M. Cui, G. Zhang, and R. Zhang, “Secure wireless communication via intelligent

reflecting surface,” IEEE Wireless Commun. Letters, vol. 8, no. 5, pp. 1410-1414,
October 2019.
 X. Guan, Q. Wu, and R. Zhang, “Intelligent reflecting surface assisted secrecy
communication: is artificial noise helpful or not?,” IEEE Wireless Communications
Letters, vol. 9, no. 6, pp. 778-782, June 2020.

IRS-aided PHY Security Rui Zhang, National University of Singapore
IRS-aided Secure Wireless Communication
 Challenging scenarios (zero secrecy rate in conventional system without IRS):

 Eavesdropping channel is stronger than legitimate channel
 Two channels are highly correlated (aligned) in space
 Objective: maximize the secrecy rate for the user via joint transmit and reflect
 Reflected signal by IRS is added constructively with non-reflected signal at the user, while
being destructively added with that at the eavesdropper
 Exploit AP's transmit beamforming to strike a balance between the signal power beamed
towards IRS and that to the user/eavesdropper for signal enhancement/cancellation

Problem Formulation and Alternating Optimization Solution

 Problem formulation
 AP transmit beamforming: w
 IRS reflect beamforming: q
Suboptimal alternating optimization method
Sub-problem 2:
Sub-problem 1:
(approximate solution
(optimal solution)
via SDR)

Secrecy Rate Performance
 The AP, the user, the eavesdropper, and the IRS are located at (0,0), (150,0), (145,0), and (145,5) in
meter, respectively. Spatial correlation matrix R, where with r =0.95
 AP transmit beamforming alone can only achieve very limited secrecy rate
 Joint design achieves constructive/destructive signal superposition at user/eavesdropper, thus
providing a new DoF to enhance secrecy rate
 With more reflecting elements, IRS beamforming becomes more flexible and achieves higher gains

IRS-aided PHY Security: Is Artificial Noise Helpful or not?
Rose
Rose
Alice Eve k Bob
 Challenge: lack of transmit DoF due to increasing number of eavesdroppers

 Conventional system without IRS: AN is helpful
 IRS-aided secrecy communication: Is AN still helpful?
 Objective: maximize the achievable secrecy rate via a joint design of
transmit/reflect beamforming with AN and investigate
 whether IRS can have any impact on the necessity of using AN
 under what conditions AN is most helpful

Problem Formulation and Solution
 Problem formulation  γ 0 ( v H H ari + h aiH ) f1 

2
 Transmit beamforming: f1 log 1 +

Ri =  , i ∈ {b, ek }
 
γ 0 ( v H H ari + h aiH ) f 2 + 1 
2
 Jamming with AN: f 2  

 IRS reflect beamforming: v max
f1 ,f2 , v
{R − max R }
b
k
ek
 IRS reflected channel: s.t. f1H f1 + f 2 H f 2 ≤ Pmax ,

H ari = diag ( h riH ) H ar vn = 1, ∀n.
 Alternating optimization
Sub-problem 1: optimizing ( f1 , f 2 ) for given v Sub-problem 2: optimizing v for given ( f1 , f 2 )
max log 1 + γ b ( f1 , f 2 )  − max log 1 + γ ek ( f1 , f 2 )  max log 1 + γ b ( v )  − max log 1 + γ ek ( v ) 
f1 ,f2 k v k
s.t. f1H f1 + f 2 H f 2 ≤ Pmax . s.t. vn = 1, ∀n.
γ 0 ( v H H ari + h aiH ) f1
2
γ 0 v H H ari f1 + h aiH f1
2
γ i ( f1 , f 2 ) ∈ {b, ek }
, i= γ i ( v) , i ∈ {b, ek }
γ 0 ( v H H ari + h aiH ) f 2 + 1
2
2
γ 0 v H ari f 2 + h f
H H
ai 2 +1
Solve the sub-problems by applying SDR with SCA.

Secrecy Rate Performance
10
z AN, IRS
9
No-AN, IRS
AN, No-IRS
8
20 No-AN, No-IRS
Setup (b)
6
Alice
Achievable Secrecy Rate (bps/Hz)

(5, 0, 20) Rose 5
Setup (a)
2
(0, 100, 2) 4
95 100 105
y 3
…… Eves 2 Eves ……
3 2
5 Bob 1
Setup (b) Setup (a)
(3, 100, 0) 0
x 15 20 25 30 35 40
P (dBm)
max
 Consider two setups, corresponding to the cases with local and remote Eves from IRS
 As Pmax increases, the AN-aided designs outperform their counterparts without AN
 Using AN is still helpful with IRS, especially for the case of local Eves (setup (a))

IRS-aided SWIPT
 Q. Wu and R. Zhang, “Weighted sum power maximization for intelligent reflecting

surface aided SWIPT,” IEEE Wireless Communications Letters, vol. 9, no. 5, pp. 586-590,
May 2020.
 Q. Wu and R. Zhang, “Joint active and passive beamforming optimization for intelligent
reflecting surface assisted SWIPT under QoS constraints,” IEEE Journal on Selected Areas
in Communications, vol. 38, no. 8, pp. 1735-1748, August 2020.

IRS-aided SWIPT Rui Zhang, National University of Singapore
IRS-aided SWIPT
AP IRS
 Information signals can be

EHR exploited for energy harvesting
 Fundamental question:
IDR
Energy charging zone
dedicated energy beamforming
or not?
Wireless control link IRS controller

Information flow Energy flow
 SWIPT bottleneck: low energy efficiency of far-field WPT

 Compensate high RF signal attenuation over long distance with IRS’s intelligent signal
reflection using a large aperture
 Create an effective energy harvesting/charging zone
 Objective: maximize the weighted sum received RF power at EHRs subject to SINR
constraints at IDRs via joint transmit and reflect beamforming optimization

Problem Formulation and Fundamental Result

 Special case: WPT only  General case: SWIPT
 Alternating optimization with SCA or SDR

 General result for SWIPT: Dedicated energy signals are not required, for
arbitrary user channels, i.e.,

IRS-aided SWIPT
-4 -5
10 10
4.5 1.4
Proposed joint design, LoS G, M=4

H
4 Strongest eigenmode of G G, LoS G, M=4
1.2
H
Strongest eigenmode of H H , LoS G, M=4
d d
3.5
Proposed joint design, Rayleigh G, M=4
H
1 Proposed joint design
Strongest eigenmode of G G, Rayleigh G, M=4
3 H Separate design with one energy beam
Strongest eigenmode of H H , Rayleigh G, M=4
d d Without IRS
Without IRS, M=4
0.8
2.5
Sum-power received at EHRs (W)
Sum-power received at EHRs (W)

Without IRS, M=20
-6
10
2
Highly correlated G improves the sum-power of EHRs 0.6 1.6
1.5 1.4
0.4
1.2
1
1
0.2 0 10 20 30
0.5
0 0
5 6 7 8 9 10 11 -5 0 5 10 15 20 25 30
The distance between the AP and the IRS, r (m) SINR target of IDRs, (dB)
0
IRS-aided WPT IRS-aided SWIPT

 Deploying IRS in line-of-sight with AP is beneficial for improving WPT efficiency
 Different from IRS deployment for information transmission, DoF vs beamforming gain
 Significantly improve the achievable power-SINR (energy-rate) region for SWIPT
 Sending dedicated energy beam suffers considerable performance loss

IRS-aided SWIPT: QoS-Constrained Beamforming Design
 Strong coupling between

optimization variables in QoS
constraints
 Many QoS constraints render
alternating optimization easily
get stuck at local optimum
 Smart IoT networks

 Establish both communication and energy hot spots by using multiple IRSs
 Guarantee QoS for both IUs and EUs
 Objective: minimize the transmit power at the AP subject to both SINR constraints at
IUs and energy harvesting constraints at EUs via joint transmit and reflect

Problem Formulation and Penalty-based Algorithm

 Optimization problem  Equivalent transformation
(1)
(2)
(3)
 New penalty-based method to

decouple QoS constraints
 Block coordinate descent
 (Semi) closed-form solutions for
s.t. (1), (2), (3) variables in each block

Simulation Setup and Convergence
 3D coordinate system
 Spherical-wave model
 IU cluster and EU cluster
 Converged solution
satisfies all the QoS
constraints

IRS-aided WPT
 As KE, i.e., number of QoS constraints
increases, proposed penalty method
achieves higher gain than alternating
optimization
 Favorable high user channel correlation
achieved by
 Tuning IRS phase shifts
 Proper IRS deployment, LoS better than Rayleigh
fading
 IRS helps
 reduce the transmit power
 reduce the number of energy beams and simplify
AP transmission
 W/O IRS
more EUs more energy beams
 W/ IRS
only one beam!
 IRS acts as a master EU

IRS-aided SWIPT
 Without IRS: dedicated energy beamforming is effective

 With IRS: dedicated energy beamforming gain is marginal (thus not needed)
 exploit information beamforming leakage to the IRS for WPT
 simplify the transmitter (energy beamforming) and receiver (energy signal
cancellation) operations for implementing dedicated energy beamforming

UAV-mounted IRS
H. Lu, Y. Zeng, S. Jin, and R. Zhang, “Aerial intelligent reflecting surface: joint placement and
passive beamforming design with 3D beam flattening,” submitted to IEEE Transactions on
Wireless Communications. https://arxiv.org/abs/2007.13295

UAV-mounted IRS Rui Zhang, National University of Singapore
Aerial IRS vs Terrestrial IRS
 Terrestrial IRS  Aerial IRS

 180 half-space reflection only  360◦ panoramic full-angle reflection
 Multiple reflections required due to NLoS  One reflection suffices due to LoS
 Constrained deployment  Flexible deployment
360◦ panoramic full-angle reflection by AIRS Reduced number of reflections by AIRS

Optimal Aerial IRS Placement
 AIRS-assisted communication
system: enabling intelligent
reflection from the sky
 Enhance signal coverage over
a given target area
 The placement of AIRS can be flexibly optimized to further improve the communication
performance
 Objective: maximize the minimum SNR within the rectangular area by jointly
optimizing the transmit beamforming of the source node, the placement and phase
shifts of the AIRS
 The objective function is the minimum SNR over a 2D area, which is difficult to be
expressed in terms of the optimization variables
 The optimization problem is highly non-convex and the optimization variables are
intricately coupled with each other in the objective function
Problem Formulation and Solution
 Problem formulation
 Transmit beamforming at the source node: 𝐯𝐯
 AIRS placement: 𝐪𝐪
 AIRS phase shifts: 𝜽𝜽
 Destination node location: 𝐰𝐰
 Two-step optimization
First step: Second step:
Concatenated
Array gain path loss
 Sub-array structure for phase shift optimization

 One-dimensional search for AIRS placement

Conclusions Rui Zhang, National University of Singapore
Conclusions
 IRS: a new and disruptive technology to achieve smart and
reconfigurable propagation environment for future wireless network
 Achieve high spectral/energy efficiency with low-cost passive

reflecting elements
 A paradigm shift of wireless communication from traditional “active

component solely” to the new “active and passive” hybrid network
 Main challenges (from the communications perspective):

 IRS deployment

Future Work Directions Rui Zhang, National University of Singapore
Promising Directions for Future Work

 IRS hardware design/prototype
 IRS reflection/channel modeling
 IRS reflection optimization for more general setups (e.g., with
partial/imperfect CSI, under hardware imperfections) and other
applications (spatial modulation, localization, etc.)
 Capacity and performance analysis of IRS-aided system/network
 Practical IRS channel estimation and low-complexity passive
beamforming designs
 IRS deployment/association/multiple access in multi-cell network
 IRS meets massive MIMO, mmWave/THz, energy harvesting, UAV,
security, wireless power transfer, etc.
 IRS integration to WiFi/Cellular
 ……

Irs Wtc2020

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Irs Wtc2020

Uploaded by

Copyright:

Available Formats

Intelligent Reflecting Surface Aided Wireless

Communication: Opportunities and Challenges

Professor, National University of Singapore

WTC Online Seminar

 Communication Design Challenges

 Conclusion and Future Work

WTC Online Seminar 2020 2

 Communication Design Challenges

 Conclusions and Future Work

WTC Online Seminar 2020 3

Have We Reached Shannon’s Capacity Limit?

Communicate with other nodes in the

 A digitally-controlled metasurface with massive low-cost passive reflecting

WTC Online Seminar 2020 5

IRS: Reflection Model

 Baseband equivalent signal model at each IRS element

where : reflection amplitude

 In practice, both amplitude and phase shift need to be discretized

IRS: Channel Model

𝑑𝑑1𝑛𝑛 element 𝑛𝑛 𝑑𝑑2𝑛𝑛

 Extendible to wide-band channel, with IRS frequency-flat reflection only

IRS Path Loss Model: Product Distance or Sum Distance?

𝑑𝑑1𝑛𝑛 element 𝑛𝑛 𝑑𝑑2𝑛𝑛

 Applies to free-space propagation and

Main Functions of IRS in Wireless Communication

WTC Online Seminar 2020 9

IRS for Channel Reconfiguration

 Create virtual LoS link by smart

WTC Online Seminar 2020 10

IRS-enabled Passive Beamforming

 3D passive beamforming for broadcasting/multicasting

IRS-assisted Interference Nulling/Cancelation

 Enhance desired signal power

WTC Online Seminar 2020 12

IRS Applications for 5G/6G

Related Industry Initiatives

WTC Online Seminar 2020 14

IRS vs Active Relay/Small Cell/DAS/Cell-free MIMO

Wireless Network with IRS vs w/o IRS: System Setup

 A fundamental question for IRS:

 New hybrid active/passive

 Analytical framework based on stochastic geometry

WTC Online Seminar 2020 16

Wireless Network with IRS vs w/o IRS: Simulation Results

 ν versus ζ under given total cost C:

 ν versus C under given density ratio ζ :

WTC Online Seminar 2020 17

IRS vs Massive MIMO/Active Large Surface

Massive MIMO IRS-aided Small MIMO

IRS vs Massive MIMO: Simulation Setup

IRS vs Massive MIMO: Simulation Results

Achievable max-min rate (bps/Hz)

AP transmit power (dBm) Delay ratio,

 M: # of active antennas at AP  This also holds considering delay due to

 Communication Design Challenges

 Conclusion and Future Work

WTC Online Seminar 2020 21

Joint Active and Passive Beamforming: Single-user Case

 AP: active (transmit) beamforming

Minimum AP transmit Power vs AP-user Distance

Problem formulation (suboptimal solutions

 Significant power saving with IRS (vs w/o IRS)

WTC Online Seminar 2020 23

Minimum AP transmit Power vs No. of IRS Elements (1)